Gigabit Ethernet works over Category 3 cable

[Eug]

Just as a lark, I bought some CAT3 to see how well Ethernet could work over CAT3. Well, to my surprise, even Gigabit worked fine over my (very) short 7-foot Cat-3 cable. iperf tells me the max speed I got was 867 Mbps, which is 1.01 Gbytes transfered in 10 seconds.

Pic

Eug-iMac-C2D:utilities eug$ ./iperf -s
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 256 KByte (default)
------------------------------------------------------------
[ 4] local 192.168.1.xx port 5001 connected with 192.168.1.xx port 60603
[ ID] Interval Transfer Bandwidth
[ 4] 0.0-10.0 sec 1.01 GBytes 867 Mbits/sec

I wonder what kind of distance once could get out of Cat-3 in an average home, at least for 100Base-TX.

 

[robmurphy]

I have seen similar when testing cabling. 100 M of Cat3 was fitted between a Cisco (3750 Cat I think) switch and a PC and the auto negotiation went to 100 meg full duplex for the link. The port on the switch was a 10/100 Meg port. I think most of the Cat3 cabling sold is basically Cat5. You will probaly find that a normal Cat5 ethernet patch lead will happlily auto negotiate to 1Gig. Over a long period You may find more transmision errors over the Cat5 lead though.

Basically the switches cannot realy tell the difference between Cat3, Cat5 and Cat5E if all 4 pairs are connected. They can tell the difference if only 2 pairs are connected (all that is needed for 10/100 Meg ethernet). Its up to the user to make sure the cable used for ethernet connections is up to the required standard.

Rob.

 

[JackMDS]

And the point is?

 

[Eug]

Originally posted by: robmurphy
I think most of the Cat3 cabling sold is basically Cat5.

I don't think that's true. CAT3 has a lot less twists. There is no mistaking CAT3 for CAT5, even just by looking at it. When you remove an inch of the outer jacket of CAT3, the twisted pairs are loose enough that it almost looks like straight wire.

Part of my house is wired with CAT5. It looks exactly the same as CAT5e, and quite different from CAT3.

You will probaly find that a normal Cat5 ethernet patch lead will happlily auto negotiate to 1Gig. Over a long period You may find more transmision errors over the Cat5 lead though.

My understanding is that the GigE spec actually was built with the original CAT5 in mind. FWIW, for the part of my house with CAT5, Gigabit Ethernet works perfectly and at full speed. However, since it's just a house the lengths are comparatively short. No 100 metre runs. The wiring is over 10 years old though.

Originally posted by: JackMDS
And the point is?

As I said in the original post, it was just as a lark. I decided to try it because someone told me it couldn't be done, and I didn't believe him. However, 7 feet isn't a very good test. I just chose 7 feet because I was too cheap to pay for a longer length of cable. However, if I can find some really inexpensive CAT3 somewhere, I'll try it with 10 metres of cable... just as a lark.

 

[DisgruntledVirus]

Welcome to the world of conservative ratings.

The Cat x cable spec says it is rated to run at this speed, and it will provided the equipment on the other ends do as well. Also, you must factor in distances. The longer the cable, the more difficult it is to maintain a high frequency signal without degredation (as you seem to know).

I'd be willing to bet that you could get gig speeds on a "Cat 3" cable on that 10m experiement as well. My guess is it's *officially* rated at Cat 3, but more likely a Cat 5 cable.

 

[ScottMac]

Originally posted by: DisgruntledVirus
Welcome to the world of conservative ratings.

The Cat x cable spec says it is rated to run at this speed, and it will provided the equipment on the other ends do as well. Also, you must factor in distances. The longer the cable, the more difficult it is to maintain a high frequency signal without degredation (as you seem to know).

I'd be willing to bet that you could get gig speeds on a "Cat 3" cable on that 10m experiement as well. My guess is it's *officially* rated at Cat 3, but more likely a Cat 5 cable.

NO. just no.

The specifications that make up the Cat5 (5e, 6, 6a) spec are far tighter than the specs required to pass Category 3.

The rating requires testing at a length of 90 meters for solid conductor. Testing speed over spec at a shorter length means that it seems to work in that instance; nothing more. It's useless for any other comparison.

We got 155 ATM to go over (literally) barbed wire, through sewing scissors, and an old brass doorknob in our Lab (at Anixter)... that doesn't mean that it has any application, in any way, shape, or form for anything other than we had some time to kill. Also getting data through, and getting *all* the data through (on the first try) are completely different animals.

This was a bullshit test and totally useless for anything beyond conversation.


Also remember that the spec requires five meters each end of jumper cordage for the total length of 100 Meters)

 

[Eug]

It's definitely not a CAT5 cable. Not enough twists. CAT5 has around 3 twists per inch. CAT3 has around 3 twists per FOOT.

Like I said, when you remove an inch of the outer jacket of this cable, it looks like straight wire, not twisted pair. To put it another way, to crimp CAT3, you don't actually have to untwist anything.

 

[Eug]

We got 155 ATM to go over (literally) barbed wire, through sewing scissors, and an old brass doorknob in our Lab (at Anixter)... that doesn't mean that it has any application, in any way, shape, or form for anything other than we had some time to kill.

Cool!

This was a bullshit test and totally useless for anything beyond conversation.

Pretty much, except in the situation where I had an extra unused CAT3 phone line in an existing house and I didn't want to drill holes, for casual networking. If it was a short run, I'd consider trying to network it through the CAT3 line. It'd be cheaper than wireless, and potentially actually more stable and faster than wireless too.

 

[ScottMac]

Originally posted by: Eug
It's definitely not a CAT5 cable. Not enough twists. CAT5 has around 3 twists per inch. CAT3 has around 3 twists per FOOT.

Like I said, when you remove an inch of the outer jacket of this cable, it looks like straight wire, not twisted pair. To put it another way, to crimp CAT3, you don't actually have to untwist anything.

You cannot judge a cable by the twists. There are many more factors that go into producing a cable that meets the higher spec. Yes, twists play some part of it, but more often the case, the difference in the twist ratio from one pair to another. There is no meaningful standard, there is no specific convention ... the producer of the cable will engineer it the way they decide to meet the published specification.

They also can change the insulating material, the lay of the pair within the sheath, the sheath material, the amount of twisting of the aggregated pair , the conductor size, thickness of the insulation.

That is why, in a serious installation, it is important to use one manufacturer from end to end for all components. They know where their weaknesses are and can counter them in another component (cable versus jack or panel). There are cases where totally compliant components can come together, and, due to collective tolerances, be out of spec in several ways.

There's more to cabling than plastic-coated copper. There's actually some serious engineering involved.

 

[Eug]

Have you ever seen any CAT5e cable that has only a couple (or three) twists per foot?

 

[JackMDS]

Let me make sure that I understand.

I have a Nice Fireplace, so if I am Short of wood or want the save few $$, I can use Horse manure(or similar substance) instead of wood to keep my house temperature Up.

Is this is the correct analogy?

 

[cmetz]

It would not surprise me if the factory that makes that cable churns out one kind of cable, and labels it cat5 or cat3 depending on demand. It's probably not even cost effective at this point to make true cat3 or cat4 anymore vs. just churning out cat5. Or perhaps the stuff that doesn't quite pass testing is labelled cat3 and harvested.

Beyond that, many technologies will mostly work in situations way outside their design spec. Ethernet, in particular, is known for mostly working in many situations it shouldn't. You decide whether you think that your network mostly working is good enough for you.

In enterprise environment, people get paid big bucks to figure out how to turn a mostly working network into a fully working network. The difference between the two can cost an enterprise a lot, and that's why they're willing to pay a lot to bridge that gap. Cut corners now, pay for it later.

 

[Eug]

Originally posted by: JackMDS
Let me make sure that I understand.

I have a Nice Fireplace, so if I am Short of wood or want the save few $$, I can use Horse manure(or similar substance) instead of wood to keep my house temperature Up.

Is this is the correct analogy?

Not really. In stores are actually firewood substitutes for fireplaces. One is pressed wood shavings, and the other is pressed left over coffee plant products. Both actually cost more than bulk firewood (but cost less when purchasing small amounts).

Your analogy is way off base anyway. It's not as if anyone is actually recommending using CAT3 for networking, when CAT5 or better is available. However, if you're stuck in a house in the middle of nowhere and the only fuel for the fireplace is dried horse manure, I'd rather use that than freeze to death.

Lemme guess. You have never tried overclocking a CPU just for fun?

Originally posted by: cmetz
It would not surprise me if the factory that makes that cable churns out one kind of cable, and labels it cat5 or cat3 depending on demand. It's probably not even cost effective at this point to make true cat3 or cat4 anymore vs. just churning out cat5. Or perhaps the stuff that doesn't quite pass testing is labelled cat3 and harvested.

Again, have you ever seen CAT5 with just a few twists per foot? I don't have much experience installing cable, but I can honestly say I have never seen any CAT5 from any era that looks like CAT3. Even back in the 90s, CAT5 (without the 'e') was much more tightly twisted than CAT3 is now.

Because CAT3 has much less twists, it also has less copper even at the same wire gauge. So even in 2009, it still makes sense to make CAT3 at less than CAT5 spec levels, for cost reasons. Remember, CAT3 is still an extremely popular standard. It is the standard of choice in fact for home installs, for phone lines. It's significantly cheaper than CAT5e, and it's a lot lighter too.

Originally posted by: ScottMac
You cannot judge a cable by the twists. There are many more factors that go into producing a cable that meets the higher spec. Yes, twists play some part of it, but more often the case, the difference in the twist ratio from one pair to another. There is no meaningful standard, there is no specific convention ... the producer of the cable will engineer it the way they decide to meet the published specification.

They also can change the insulating material, the lay of the pair within the sheath, the sheath material, the amount of twisting of the aggregated pair , the conductor size, thickness of the insulation.

That makes sense, but it still comes down to this... Have you ever seen CAT5 that looks like conventional CAT3, with just a few twists per foot?

 

[ScottMac]

Originally posted by: Eug

Originally posted by: JackMDS
Let me make sure that I understand.

I have a Nice Fireplace, so if I am Short of wood or want the save few $$, I can use Horse manure(or similar substance) instead of wood to keep my house temperature Up.

Is this is the correct analogy?

Not really. In stores are actually firewood substitutes for fireplaces. One is pressed wood shavings, and the other is pressed left over coffee plant products. Both actually cost more than bulk firewood (but cost less when purchasing small amounts).

Your analogy is way off base anyway. It's not as if anyone is actually recommending using CAT3 for networking, when CAT5 or better is available. However, if you're stuck in a house in the middle of nowhere and the only fuel for the fireplace is dried horse manure, I'd rather use that than freeze to death.

Lemme guess. You have never tried overclocking a CPU just for fun?

Originally posted by: cmetz
It would not surprise me if the factory that makes that cable churns out one kind of cable, and labels it cat5 or cat3 depending on demand. It's probably not even cost effective at this point to make true cat3 or cat4 anymore vs. just churning out cat5. Or perhaps the stuff that doesn't quite pass testing is labelled cat3 and harvested.

Again, have you ever seen CAT5 with just a few twists per foot? I don't have much experience installing cable, but I can honestly say I have never seen any CAT5 from any era that looks like CAT3. Even back in the 90s, CAT5 (without the 'e') was much more tightly twisted than CAT3 is now.

Because CAT3 has much less twists, it also has less copper even at the same wire gauge. So even in 2009, it still makes sense to make CAT3 at less than CAT5 spec levels, for cost reasons. Remember, CAT3 is still an extremely popular standard. It is the standard of choice in fact for home installs, for phone lines. It's significantly cheaper than CAT5e, and it's a lot lighter too.

Originally posted by: ScottMac
You cannot judge a cable by the twists. There are many more factors that go into producing a cable that meets the higher spec. Yes, twists play some part of it, but more often the case, the difference in the twist ratio from one pair to another. There is no meaningful standard, there is no specific convention ... the producer of the cable will engineer it the way they decide to meet the published specification.

They also can change the insulating material, the lay of the pair within the sheath, the sheath material, the amount of twisting of the aggregated pair , the conductor size, thickness of the insulation.

That makes sense, but it still comes down to this... Have you ever seen CAT5 that looks like conventional CAT3, with just a few twists per foot?

Yes, I have. Which pair are you talking about? They are not all twisted at the same rate. Look at the (frequently) brown pair ... lightly twisted ... check that against the (usually) green or orange pairs ... more tightly twisted, relative to the other pairs.

Some of the flat / under-carpet Cat5 cables are hardly twisted at all.

Also note that not all Unshielded Twisted Pair cabling has any Category rating at all. Plain old phone wire is not rated (quad, DIW ...) at all ... and some of that is twisted too, but there's no effort to make it meet Cat specs, because they don't have to.

Cabling producers make the cable in batches. One batch is Cat5e, the next batch might be Cat6/6a, the next batch might be quad/DIW, the next batch might be Cat3 ... they intentionally make it to the spec they're making in that batch. Companies cannot afford to crap shoot on what might roll out of the machines, they get what they make. Anixter, as part of their quality process, randomly tests each batch coming into their warehouses for Category compliance. If the sample doesn't pass, the entire shipment goes back to the manufacturer. I used to work in that Lab (I didn't do cable, I was one of the Interoperability folks).

There's more to Category rating that twists.

Over-twisting the pair has consequences too. You can't uprate a cable just by adding more twists.

 

[Eug]

Originally posted by: ScottMac
Yes, I have. Which pair are you talking about? They are not all twisted at the same rate. Look at the (frequently) brown pair ... lightly twisted ... check that against the (usually) green or orange pairs ... more tightly twisted, relative to the other pairs.

What's the average ballpark twist rate for these?

As for the brown pair, all of the bulk CAT5e I've seen (which is obviously much less than what you've seen) has always had more than 1 twist per inch, whereas it's several fold less twists for CAT3, even in 2009. Like I said before, most CAT3 can almost look like straight wire at times.

Some of the flat / under-carpet Cat5 cables are hardly twisted at all.

Good point. I've always wondered about those though. Do they sell it in bulk, or are they only patch cables? ie. Can you run 100 metres of flat (nearly untwisted pair) CAT5?

Also note that not all Unshielded Twisted Pair cabling has any Category rating at all. Plain old phone wire is not rated (quad, DIW ...) at all ... and some of that is twisted too, but there's no effort to make it meet Cat specs, because they don't have to.

Yeah, but we're specifically talking about CAT3 and CAT5 here, not unrated cable. As for home builds here with CAT3 though, it's usually not 4-pair. It's usually 2-pair or 3-pair.

There's more to Category rating that twists.

Of course there is. However, it seems to be a significant component. You are arguing absolutes. I agree in that sense. The spec does not specify design parameters. It just specifies specific performance parameters as I understand it.

Nonetheless, mainstream companies selling bulk CAT3 and bulk CAT5e cable have designs that are significantly different for the two cable categories, which are also different enough that you can tell them apart just by looking at them. To put it another way, sell a roll of traditional mainstream CAT3 to someone wanting to set up a 1000 Mbps network, and just by looking at the wires, they'll instantly know you're trying to rip them off, without having to look at the label. This is definitely not true for CAT5 vs. CAT5e. They pretty much look identical to someone like me.

If it wasn't preferable to add these twists for CAT5e, I suspect nearly all CAT5e would get rid of the twists, because it costs more in terms of manufacturing and for materials (copper).

 

[Eug]

So FWIW, my 5-foot flat Ethernet cable arrived today. It's not rated as anything and consists only of 8 parallel (stranded) wires embedded in a common flat plastic sheath, with each wire about 1 mm apart from the next. The cable is terminated on each end with an RJ45 connector. On a Gigabit network, I get only 93 Mbps.

 

[Colt45]

I wonder what speed you can get out of coathangers.

 

[Eug]

Heh. Scott got 155 ATM to work through a door knob.

I'm sure coathangers would work fine, although one wonders about 7 feet of coathangers.

 

[gmelander1]

I just had a class run a speed test over 100 feet of CAT3 cable. They were able to achieve the same speed (about 500 Mbps) using a gigabit switch and gigabit LAN cards as they achieved with CAT6 cable.

Now, in doing this test, there was only the one cable (actually 2, since there were two computers and one switch involved). My theory is that if we wired the entire LAN with CAT3 cable and all of the computers were actively sending/receiving data, that there would be so much crosstalk/interference that no significant data would successfully pass.

Has anyone tested in this environment? I know from field experience that when a company upgraded their switch without replacing their CAT3 cabling, it performed VERY poorly.

 

[BrianB_NY]

I realize that this is an old thread, but feel like I need to correct some things for others who come across this thread.

There are a couple of factors that determine how well a transmission medium will work for moving an electronic signal.
1. Characteristic Impedance
2. Crosstalk

Characteristic impedance is the AC "resistance" of the medium (copper wire in this case). 10/100/1000 802.3 Ethernet is all designed for 100-ohm differential impedance. Any two set of conductors that are situated in such a way as to provide this impedance will carry the signal just fine [1], at least in one direction. That's where the second part comes in...

Crosstalk in the conduction of one signal on a wire to an adjacent wire due to the electromagnetic (EM) fields generated by the moving electrons. This is why the pairs of wires are twisted. The differential signal (same transitions but opposite polarity) on the twisted wires help cancel out the EM fields so that they are picked up by the adjacent pairs in the cable. Crosstalk causes distortion of the signal that the victim wire is trying to carry.

So.... if the impedance is close enough and the crosstalk is small enough then you're likely to get a working connection between the two ends.

Category-3 cable is designed for 10Mbps. By itself it may perform better. However in an installation where there are multiple of these cables bundled (such as entering a switch/router rack) it's unlikely to do better than the 10Mb.

I also saw someone mention about different number of twists between pairs in a cable. This is done so that the twists don't line up between pairs. This makes sure the EM fields don't amplify.

[1] In Dr Howard Johnson's book "High-Speed Digital Design, The Art of Black Magic" he tells a story about a booth at a conference where a manufacturer of ethernet PHY ICs was showing how superior their chips were because they worked over barbed wire. He pulled out a ruler and measured the distance between the conductors and determined that they were spaced perfectly to provide the correct impedance. Any manufacturers chips would have worked with that setup.

 

[MtnMan]

Customer called that new PC would not connect to the network in the location where an old PC had been working for years. But if they took the new PC to a different location, it connected fine. After listening to the companies "PC guy" explain all he had done, I put a cable tester on the drop. The drop was almost 150 meters in length, but the old PC's NIC was perfectly happy with that and had worked for years. New PC NIC, was 'sorry don't think so'. This was back in the CAT 3 days.

The guy ended up swapping the NIC out of the old PC into the new one and disabling the onboard NIC.

I wrote them a bill for 2 hours.

 

[DaaQ]

I had only ever seen CAT3 as 2 pair. Typically to run two telephone lines to one location.

Have never bought into the flat cable "phenomenon", nor CAT7 + meaning CAT 7 or higher for patch cables. Like Monster ect.
It is like the "gold" rg6 F fittings. Must be better right? Screw on, Crimp, or Compression. IDK Wallyworld or Geeksquad knows more right. RIGHT.
Am I right?










/s just in case.